8192 is very precise, precisely 2^13

--- In protomuck@yahoogroups.com, Kelketek Rritaa <kelketek@...> wrote:
>
> 8192? That's an awfully precise number.
> -----Original Message-----
> From: Alynna Trypnotk <alynna@...>
>
> Date: Fri, 13 Feb 2009 10:07:43
> To: <protomuck@yahoogroups.com>
> Subject: re: [Protomuck] Speed up MUF execution
>
>
>
> > Wow, then 100 really is low! Guess I'll crank that sucker up.
> >
> > ------------------------------------------------------------------
------
> > From:
> > Date: Mon, 9 Feb 2009 11:21:56 -0600
> > To: <protomuck@yahoogroups.com>
> > Subject: Re: [protomuck] Speed up MUF execution
> >
> > Hm, I think the default is usually 10,000 .... I believe mine are set to
> > 84,000. Try different values, you should be able to figure out where the
> > best spot is.
> >
> > -Michael
> > ----- Original Message -----
> > From: "Kelketek Rritaa" <kelketek@...
> <mailto:kelketek%40gmail.com>>
> > To: <protomuck@yahoogroups.com <mailto:protomuck%40yahoogroups.com>>
> > Sent: Monday, February 09, 2009 11:11 AM
> > Subject: Re: [protomuck] Speed up MUF execution
> >
> > > Ain't my program :-P These ones are found in the standard DB. but only
> > > 100 instructions per program default? I'd think my machine is fast
> > > enough to multiplex well enough with at least 2000. Probably more.
> > >
> > > Thanks!
> > >
> > > On Feb 9, 2009, at 11:04 AM, <hinoserm@...
> > <mailto:hinoserm%40protomuck.org>> wrote:
> > >
> > >> You may want to raise your instr_slice @tune, although at the point
> > >> where
> > >> the default is introducing delays you may want to rethink the
> > >> program design
> > >> first.
> > >>
> > >> -Michael
> > >> ----- Original Message -----
> > >> From: "Jonathan Piacenti" <kelketek@...
> > <mailto:kelketek%40gmail.com>>
> > >> To: <protomuck@yahoogroups.com <mailto:protomuck%40yahoogroups.com>>
> > >> Sent: Monday, February 09, 2009 10:18 AM
> > >> Subject: [protomuck] Speed up MUF execution
> > >>
> > >> Is there any way to speed up the execution of MUF programs? I'm using
> > >> a say replacement program and it's inducing a three second lag.
> > >> There's more than enough processing power on the machine to ensure
> > >> instantaneous execution, so what's the deal?
> >
> > instr_slice lower than 1000 is actually very bad, it doesn't actually
> > have anything to do with the speed the MUF executes itself, but it
> > does have everything to do with how often the MUF interpreter task
> > switches.
> >
> > at 100 the MUF interpreter will be spending about as much time
> > switching between MUF programs as it does executing them.
> >
> > A good value for this tune is 8192. Lower numbers switch tasks more
> > often, which will affect the total speed of the programs running, in
> > favor of giving tasks time more often. However at instr_slice=100,
> > about as much time will be spent in the scheduler as in the MUF
> > interpreter itself. However very HIGH numbers (like 65000) will cause
> > MUF programs to become less and less fair about their time, another
> > words it could spend a very long time in a single program and other
> > programs will get slower and experience delays..
> >
>